This invention relates generally to characterizing a collection of content items, and more particularly to categorizing content items in the collection based on the address information associated with content items.
A digital magazine identifies content items for presentation to users based on user-defined preferences or parameters. These content items are presented to users via devices such as mobile devices, tablet computers, laptop computers, or desktop computers. As a user interacts with content items presented in a digital magazine, additional content items may be identified for presentation to the user based on the interactions. For example, the additional content items are selected based on their similarity to content items with which the user previously interacted.
Conventional methods for determining similarity of various content items compare content presented by the content items. For example, to compare textual articles, comparing the frequency with which keywords appear in different articles provides a measurement of similarity. However, if content items include a variety of types of content, such as image data, audio data, video data, text data, advertisements, or other content, conventional methods are less capable of accurately determining similarity between different content items. When presenting content in a digital magazine, identifying content items similar to those previously presented to a user is likely to increase user interaction with the digital magazine.